[Theme] Continuous identification of an AI Agent
fg-tida-themes-theme-continuous-identification-of-an-ai-agent-16
| Type | FG-TIDA document |
|---|---|
| Group | Trust and Identity for Humans and Agentic AI |
| State | Submitted |
| Authors | Not entered |
| Last updated | 2026-10-23 15:10 UTC |
Formats and resources
Abstract
Theme Name
During its life, an AI Agent may evolve. These evolutions may affect the behaviour, capacity and actions of an AI Agent. For traceability reasons it is important to identify these exact aspects of an AI Agent (1) which may evolve over time, and (2) affect the behaviour, capacity and actions of an AI Agent, comprising its “configuration”.
Proposer(s)
Alban FERAUD - IN Groupe - alban.feraud@ingroupe.com
Summary
During its life, an AI Agent may evolve as a result of:
• continual learning from the inputs and situation it faced;
• New connected tools;
• Etc.
These evolutions may affect the behaviour, capacity and actions of an AI Agent.
Such evolving AI Agent may be used for various purposes or in various use case such as fraud detection (learning from past analysis and/or using new tools), etc…
For traceability reasons it is important to identify these exact aspects of an AI Agent (1) which may evolve over time, and (2) affect the behaviour, capacity and actions of an AI Agent, comprising its “configuration”. In addition, this identification should be deterministic, meaning that the identification of the same AI Agent obtained at various moment in time or using different tools shall be the same. Establishing such a traceability is key to:
• Ensure accountability of the AI Agent owner and the AI Agent supplier;
• support forensic investigation in case of breach or violation;
Scope / Open Questions
As part of this theme, the following aspects could be further analysed:
• Listing of the aspects of an AI Agent likely to evolve over time and affecting its behaviour, capacity and actions, comprising its “configuration”;
• Method to uniquely identify in a deterministic manner such “configuration”;
• Internal services (to get the identification of the “configuration”) and external interfaces to be provided by an AI Agent to provide such identification prior to carrying out an operation;
Related Work
To be identified
Related Themes
To be identified
Discussion
Revisions
- No file revisions have been uploaded.